Analyzing Non-functional Requirements of Embedded Systems

D. Monjau and M. Sporer (Germany)


Modelling, Estimation, Robotics


This contribution describes a method for the static/dynamic analysis of non-functional requirements of embedded systems. Embedded systems consist of a varying number of processing- and communication functions. They are implemented by microprocessors, memory units, buses, sensors, actuators, etc. Not only is the correct operation of these components required, but the fulfilment of non-functional orders also (execution-time of a function, energy-consumption, area of the chip, and the costs and reliability of the overall system). Since there is already a great number of possible implementations, the system engineer should be supported in the selection of the components. Our system’s support contains the selection of suitable components and the analysis of the qualities which a system consisting of these components will have. The method is characterised by the strict reuse of specifications, functions, prototypes, architectures, and component implementations for hardware/software systems. The development-engineer specifies the new system’s functional and non-functional requirements on the system level. A prototype, solution-architectures and implementations are automatically generated by our design tools. In this context static and dynamic analyses of non-functional characteristics are necessary. A corresponding tool analyses the dynamic timing behaviour and the capacity of the system under construction and its components with the intention of measuring both response-times and execution- and waiting-times of processes. This is the basis for the estimation of throughput rates, bottlenecks, timing capacity of the components and further characteristics of the performance of an embedded system under workload conditions. This approach is illustrated by functions of the domain of robotic controls [3].

Important Links:

Go Back